|Publication number||US7392238 B1|
|Application number||US 10/018,518|
|Publication date||Jun 24, 2008|
|Filing date||Aug 23, 2000|
|Priority date||Aug 23, 2000|
|Also published as||WO2002027541A1|
|Publication number||018518, 10018518, PCT/2000/246, PCT/CN/0/000246, PCT/CN/0/00246, PCT/CN/2000/000246, PCT/CN/2000/00246, PCT/CN0/000246, PCT/CN0/00246, PCT/CN0000246, PCT/CN000246, PCT/CN2000/000246, PCT/CN2000/00246, PCT/CN2000000246, PCT/CN200000246, US 7392238 B1, US 7392238B1, US-B1-7392238, US7392238 B1, US7392238B1|
|Inventors||Joe F. Zhou, Weiquan Liu|
|Original Assignee||Intel Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (18), Non-Patent Citations (1), Referenced by (26), Classifications (11), Legal Events (5)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The invention relates to searching across a network. More specifically, the invention relates to a method and apparatus for a concept-based search across a number of web sites within a network.
With the advent of the Internet and the World Wide Web (WWW), the vast amount of information available across the different web sites on the Internet continues to grow as the number of networks and servers connected thereto continue to increase on a global scale. Accordingly, the availability of this information on the Internet enables users to obtain such information quickly. However, locating such information on the Internet can sometimes be difficult because of the vast numbers of websites available.
Currently, particular web sites on the WWW include search mechanisms termed “search engines” that have been developed in order to assist the users of the WWW in locating specific information therein. In particular, software programs, termed crawlers or spiders, automatically traverse, on a periodic basis, the WWW through the different web sites located thereon based on hyperlinks among the different web sites. When these software programs locate a web site whose information is not stored in the local database for this particular search engine, such programs transmit keywords and other relevant information concerning the new web site back to the web site executing these programs. Accordingly, this information about the new websites is stored into the local database for the search engine website.
Moreover, because of the dynamic nature of the WWW, these software programs also update such relevant information concerning a web site whose information is already stored in the search engine's local database. Accordingly, these search engine web sites store these search terms in a local database with other search terms from other websites. In operation, a user of these search engines enters a search query into the search engine and based on this local database, the search engine locates web pages on different web sites whose search terms include the search query and retrieves portions of these web pages. The search engine then displays these portions of the web pages to the user, thereby allowing the user to select from among the different web pages.
Additionally, another type of search engine for the WWW performs a search among the search engine results described above. In particular, these type of search engines perform a search among the individual search engines as described above and combines the results from these individual search engines into a combined results, thereby allowing the user to obtain results from across multiple search engines for a single search query.
Embodiments of the invention may be best understood by referring to the following description and accompanying drawings which illustrate such embodiments. In the drawings:
In the following description, numerous specific details are set forth to provide a thorough understanding of the invention. However, it is understood that the invention may be practiced without these specific details. In other instances, well-known structures and techniques have not been shown in detail in order not to obscure embodiments of the present invention.
While different embodiments of the present invention could include different types of networks, in an embodiment, network 102 is the Internet that includes the WWW. However, embodiments of the present invention could include other types of networks. For example, in one embodiment, network 102 is a local area network (LAN). In another embodiment, network 106 is a wide area network (WAN). Further, network 106 can be a combination of different networks that provide communication among servers 104 a–i and clients 106 a–i. In operation for a given client, such as client 106 a, connecting to a given server, such as server 104 a, client 106 a connects to server 104 a through network 102 and is able to download and upload different web pages from and to different websites residing on server 104 a.
Database 208 and concept networks 206 are associated with website 220, which is stored on servers 104. In particular, website 220 acts as a search engine that enables clients 106 to search web pages on the local server, which stores this website, as well as remote servers coupled to network 102. In particular, website 220 is a search engine that allows the users thereof to enter search terms on a given web page and have this search engine return the results of a search. Additionally, concept networks 206 are defined to include one to any number of networks, whose creation is described in more detail below in conjunction with
Moreover, servers 104 include query reformation unit 210, search engine unit 212, summarizer 214, network generation unit 216 and concordance tool 218, which, in one embodiment, are software programs that reside in memory 202 and processing unit 204 during their execution on processing unit 204. Query reformation unit 210, search engine unit 212, summarizer 214, network generation unit 216 and concordance tool 218 will be described in more detail below. In conjunction with system 100 of
Upon receiving this initial search query, server 104 reformulates this query, at process block 304. In particular, server 104 searches a network within concept networks 206 to find terms related to the initial search term or phrase.
Moreover as shown in
“Hyponym of” is another example of a semantic relationship. A hyponym relationship denotes a relationship of a subcategory to a more general class. An illustration of a hyponym is a chair being a subcategory of furniture. Moreover, another semantic relationship is a hypernym one, which is a relationship of a general class to a subcategory. For example, automobile is a hypernym of a car. Further, another semantic relationship between the terms within the network is a “part” relationship. For example a between the terms within the network is a “part” relationship. For example a bicycle tire is part of a bicycle. Accordingly, each link with concept networks 206 includes a description of the semantic relationship between the two terms. Moreover, the above-described semantic relationships are by way of example and not by way of limitation as other semantic relationships can be formed between the key term and the related terms within concept networks 206.
Method 600 of
Upon retrieving the relevant web pages, network generation unit 216 identifies the key terms in the relevant web pages, at process block 604. In an embodiment, prior to identifying the key terms, network generation unit 216 filters out “noise” words from these relevant web pages. “Noise words include common articles, conjunctions and/or prepositions. Examples of common articles include “a”, “an” and “the”, while examples of conjunctions include “and”, “or” and “nor”. Moreover, examples of prepositions include “of”, “in” and “for”. However, “noise” words are not limited to those described above, as any other words that do not carry meaning in terms of the content of the web page can be so classified.
In one embodiment, network generation unit 216 identifies these key terms employing a statistical frequency analysis of the relevant web pages. In particular, network generation unit 216 performs an analysis for each of the relevant web pages to determine the frequency of the different words and/or phrases therein. Accordingly, network generation unit 216 labels the key terms of these relevant web pages as those which occur most frequently therein. In one embodiment, the number of key terms to be identified is predetermined. For example, network generation unit 216 could identify the ten most frequent terms and/or phrase. However, embodiments of the invention are not so limited, as other techniques can be employed to determine whether a terms and/or phrase occurs frequently enough to be denoted as “key”. For example, in an embodiment, the key terms and/or phrases are those which occur “x” number of times within the given web page.
In one embodiment, network generation unit 216 identifies these key terms employing a weighting algorithm with regard to the relationship of two terms within a given web page. In one such embodiment, a weight related to the mutuality between two words is assigned to the terms of the given web page based on equation #1 shown below:
MI(x,y)=f(x,y)/f(x)+f(y)−f(x,y) (Equation # 1)
Mutual Information (MI) for two words (x and y) define the mutuality between x and y, as is known in the art. In particular, f(x) is the occurrence frequency of word x, while f(y) is the occurrence frequency of word y. Moreover, f(x,y) is the occurrence frequency of the word pair (x,y). The higher the value of MI(x,y) is, the more likely x and y are to form a compound phrase (i.e., the key phrase) within a given web page or document.
Accordingly, network generation unit 216 identifies these key terms based on the weights related to mutuality assigned by equation #1 to the terms within the given web page. In one embodiment, network generation unit 216 employs both the statistical frequency analysis and the weighting algorithm together to determine the key terms within each of the relevant web pages. These identified key terms are also known as anchor terms for their subsequent incorporation into the network of concept networks 206.
Upon identification of the key or anchor terms for a given subject matter domain, network generation unit 216 passes these terms into concordance tool 218. Accordingly, a data or text analyst uses concordance tool 218 to identify related terms to these anchor terms. In one embodiment, data analyst in conjunction with concordance tool 218 performs an analysis for a given key or anchor term separate from an analysis of other such terms. In particular, concordance tool 218 retrieves lines from the relevant web pages that include the key or anchor term, at process block 606. Such lines can be sentences or portions of sentences that include the key or anchor term from the relevant web pages. Additionally, these lines are displayed to a data analyst, who can select terms related to the key or anchor term, at process block 608.
Additionally, in one embodiment, concordance tool 218 arranges lines 702 a–702 i in alphabetical order according to the word just prior to the anchor term, as illustrated in display 700. In an alternative embodiment, concordance tool 218 arranges lines 702 a–702 i in alphabetical order according to the word just subsequent to the anchor term. These above-described orderings of lines 702 a–702 i are by way of example and not by way of limitation, as embodiments of the present invention can provide other ordering techniques, such as a random order.
Upon displaying of the lines that include the anchor terms, the data analyst employing concordance tool 218 can select terms related to these anchor terms. Returning to display 700, a given data analyst could select “baseball”, “rubber-soled” and “spikes” from line 702 a as being related terms. Moreover, in one embodiment, the data analyst defines a semantic relationship for each related term and the particular anchor term. Examples of such semantic relationships are described above in conjunction with
Returning to process block 304 of
However, embodiments of the present invention are not so limited, as any other search operator can be incorporated into the reformulated search query at process block 304. For example, other search operators to connect the initial search term with the related search terms could include OR, NOT, NEAR, etc. The search operator “OR” provides a search wherein a match is accorded when either the search term or any of the related terms are located in a given web page.
With regard to the “NOT” search operator, this operator enables a search wherein a match is accorded when a search term or related search term is specifically not found in a particular web page. Additionally, the search operator “NEAR” allows for a search wherein a match is accorded when two terms are within “x” number of words of each other in the web page. Moreover, different search operators can be combined and intermixed together in the generation of a reformulated search query. For example, a combined search operator query could be: “INTELŪ near PENTIUMŪ and INTC”. Accordingly, a web page is considered a match when INTELŪ is within “x” number of words of each other and when INTC is not found in the web page. The prior examples of search operators and combinations thereof are by way of example and not by way of limitation, as other types of search operators as well as different combinations of search operators could be included in the reformulated search queries.
In an embodiment, in addition to entering a search term or phrase in box 404 of web page 402, web page 402 could include a selection criteria, such as a radio box, which enable client 106 to indicate the level of detail and/or number of web pages to be returned from the search query. For example, a radio box could be included in web page 402 that enables client 106 to enter one of three options for level of detail: (1) high, (2) medium or (3) low. Accordingly, this option allows client 106 to control the number of web pages returned by the search result. In particular, if the level of detail is low, server 104 could incorporate more AND search operators into the reformulated search query in order to narrow the search. Conversely, if client 106 wants a high level of detail, server 104 could incorporate more OR search operators into the reformulated search query in order to expand the search.
In one embodiment, server 104 generates the reformulated search query based on a set of parameters. For example, server 104 could be designed to incorporate all related terms into the reformulated search query using an OR search operator. In further illustration, server 104 could be designed to incorporate all related items employing an AND search operator.
Moreover, in an embodiment, server 104 generates the reformulated search query based on the type of semantic relationship between the search term and the related term. In particular, web site 220 may be set up to incorporate the related terms that have certain types of semantic relationships with the search term. For example, only the related terms that have a semantic relationship of “product of” may be incorporated into the reformulated search query. The above-described examples and illustration of the reformulation of the search query at process block 304 are by way of illustration and not by way of limitation, as other types methods may be employed in the reformulation of this search query.
Returning to method 300, once the search query has been reformulated, database 208 is searched to locate the web pages that match the terms in the reformulated search query at process block 306. In particular, database 208 includes key data terms and phrases from particular documents from web pages across network 102. In one embodiment, database 208 is generated using software programs, termed crawlers or spiders, that automatically traverse, on a periodic basis, the WWW through the different web sites thereon based on hyperlinks among the different web sites. When these software programs locate a web site whose information is not stored in database 208 for this particular search engine, such programs transmit key data terms/phrases and other relevant information concerning the documents found on the new web site back to the web site executing these programs. Accordingly, this information about the new websites is stored into database 208 for website 220.
Upon locating the relevant documents from the web pages that have terms that match the terms in the reformulated search query, search engine 212 retrieves “N” number of such documents for display to the user of web site 220. In one embodiment, the user of web site 220 defines “N”. Moreover, in an embodiment, search engine 212 orders such documents for display based on relevancy to the reformulated search query. In one such embodiment, relevancy is based on how many times the search terms and related terms in the reformulated search query are found in a given document.
Additionally, summarizer unit 214 generates a summary for each of these documents. In one embodiment, the summary is the leading text of a given document. For example, the summary could be the first 500–100 words of the particular document. In an alternative embodiment, summarizer unit 214 generates the summary based on key sentences within the given document. In one such embodiment, the weight of each sentence within a given document is calculated based on how frequent the search terms within the query are found in the sentence. Accordingly, the more frequent the search terms are found in a sentence, the more weight is given to that sentence.
Summarizer unit 214 ranks the sentences based on these calculated weights, with the higher rank being assigned to the sentences with greater weights. In one embodiment, the summary is, for example, the first five highest ranked sentences. However, embodiments of the present invention are not so limited, as other techniques can be employed to determine the summary. For example, the summary could be the first 50–100 words based on the ranking of the sentences. Accordingly, if the first two highest ranked sentences include 100 words, then these two sentences define the summary.
In an embodiment, in addition to this summary, the headlines or titles for the documents are also displayed at process block 308. In one embodiment, the key words and the related words are also displayed at process block 308. This user of search engine 212 has an opportunity to select one of the related terms, at process decision block 310. If the user is satisfied with the search and the results thereof, the process is complete, at process block 314.
However, if the user wants to find different documents based on a new search, they can select one of the related terms for a new query. Search engine 212 receives this selection of a new query, at process block 312. Search engine 212 reformulates a new search query based on this new term, at process block 304. Accordingly, web site 220 performs a recursive process of receiving a new search term, reformulating the query and displaying the search results until the user is satisfied with their search. Further, in an embodiment, differing weights are assigned to the search terms and the related terms in order to place different emphasis on such terms during the search. In one such embodiment, search engine 212 places greater weight on those related terms that the user selected during the recursive searching process. For example, a related term, which the user selected, may only need to be found in a given document once, while the other terms in the reformulated search query may be required to be found multiple times in order for a given document to be considered a match. However, embodiments of the present invention are not so limited as other techniques may be incorporated into embodiments of the present invention to provide for differing weights among the different terms of the reformulated search query.
In one embodiment, web site 220 includes conversion unit 219 Conversion unit 219 allows the user to not only read, but also hear, the search results, as conversion unit 219 converts the text of the search results to speech. Accordingly, any type of user, including a novice, can use embodiments of web site 220 to search across vast numbers of web sites located on servers 104 across network 102 to find particular information in documents therein.
Moreover, the above-description was described in terms of the Internet and the WWW. However, this description was by way of example and not by way of limitation, as embodiments of the present invention are applicable to any other type of network that includes server-type computers that store data information that can be retrieved by client computers. For example, in one embodiment, network 102 could be a local area network with multiple servers and clients coupled thereto.
Memory 202 includes a machine-readable medium on which is stored a set of instructions (i.e., software) embodying any one, or all, of the methodologies described above. Software can reside, completely or at least partially, within memory 202 and/or within processing unit 204. For the purposes of this specification, the term “machine-readable medium” shall be taken to include any mechanism that provides (i.e., stores and/or transmits) information in a form readable by a machine (e.g., a computer). For example, a machine-readable medium includes read only memory (ROM); random access memory (RAM); magnetic disk storage media; optical storage media; flash memory devices; electrical, optical, acoustical or other form of propagated signals (e.g., carrier waves, infrared signals, digital signals, etc.); etc.
Although the present invention has been described with reference to specific exemplary embodiments, it will be evident that various modifications and changes may be made to these embodiments without departing from the broader spirit and scope of the invention. For example, the above-described embodiments were described in terms of the English language. However, such embodiments can be incorporated and be applicable to any other language. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5864871 *||Jan 28, 1997||Jan 26, 1999||Multex Systems||Information delivery system and method including on-line entitlements|
|US5926811 *||Mar 15, 1996||Jul 20, 1999||Lexis-Nexis||Statistical thesaurus, method of forming same, and use thereof in query expansion in automated text searching|
|US5933822 *||Jul 22, 1997||Aug 3, 1999||Microsoft Corporation||Apparatus and methods for an information retrieval system that employs natural language processing of search results to improve overall precision|
|US5987460 *||Jul 3, 1997||Nov 16, 1999||Hitachi, Ltd.||Document retrieval-assisting method and system for the same and document retrieval service using the same with document frequency and term frequency|
|US6006225 *||Sep 1, 1998||Dec 21, 1999||Amazon.Com||Refining search queries by the suggestion of correlated terms from prior searches|
|US6026388 *||Aug 14, 1996||Feb 15, 2000||Textwise, Llc||User interface and other enhancements for natural language information retrieval system and method|
|US6028605 *||Feb 3, 1998||Feb 22, 2000||Documentum, Inc.||Multi-dimensional analysis of objects by manipulating discovered semantic properties|
|US6038560 *||May 21, 1997||Mar 14, 2000||Oracle Corporation||Concept knowledge base search and retrieval system|
|US6094649 *||Dec 22, 1997||Jul 25, 2000||Partnet, Inc.||Keyword searches of structured databases|
|US6094652 *||Jun 10, 1998||Jul 25, 2000||Oracle Corporation||Hierarchical query feedback in an information retrieval system|
|US6101491 *||Mar 31, 1997||Aug 8, 2000||Sun Microsystems, Inc.||Method and apparatus for distributed indexing and retrieval|
|US6101503 *||Mar 2, 1998||Aug 8, 2000||International Business Machines Corp.||Active markup--a system and method for navigating through text collections|
|US6163778 *||Feb 6, 1998||Dec 19, 2000||Sun Microsystems, Inc.||Probabilistic web link viability marker and web page ratings|
|US6169986 *||Oct 1, 1999||Jan 2, 2001||Amazon.Com, Inc.||System and method for refining search queries|
|US6460029 *||Dec 23, 1998||Oct 1, 2002||Microsoft Corporation||System for improving search text|
|US6510406 *||Mar 22, 2000||Jan 21, 2003||Mathsoft, Inc.||Inverse inference engine for high performance web search|
|US6665665 *||Jul 30, 1999||Dec 16, 2003||Verizon Laboratories Inc.||Compressed document surrogates|
|US20020059161 *||Nov 3, 1998||May 16, 2002||Wen-Syan Li||Supporting web-query expansion efficiently using multi-granularity indexing and query processing|
|1||*||http://web.archive.org/web/20000815211428/http://www.monash.com/spidap4.html. Linda Barlow, copyright, 1996-1999. All rights reserved. Updated Apr. 11, 2000.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7685118 *||Aug 12, 2004||Mar 23, 2010||Iwint International Holdings Inc.||Method using ontology and user query processing to solve inventor problems and user problems|
|US8250053||Feb 24, 2009||Aug 21, 2012||Microsoft Corporation||Intelligent enhancement of a search result snippet|
|US8266185||Oct 21, 2009||Sep 11, 2012||Cortica Ltd.||System and methods thereof for generation of searchable structures respective of multimedia data content|
|US8280877 *||Sep 21, 2007||Oct 2, 2012||Microsoft Corporation||Diverse topic phrase extraction|
|US8818916||Jun 23, 2010||Aug 26, 2014||Cortica, Ltd.||System and method for linking multimedia data elements to web pages|
|US8868619||Sep 4, 2012||Oct 21, 2014||Cortica, Ltd.||System and methods thereof for generation of searchable structures respective of multimedia data content|
|US9031999||Feb 13, 2013||May 12, 2015||Cortica, Ltd.||System and methods for generation of a concept based database|
|US9256668||Aug 29, 2013||Feb 9, 2016||Cortica, Ltd.||System and method of detecting common patterns within unstructured data elements retrieved from big data sources|
|US9372940||Aug 29, 2013||Jun 21, 2016||Cortica, Ltd.||Apparatus and method for determining user attention using a deep-content-classification (DCC) system|
|US9396435||Oct 8, 2014||Jul 19, 2016||Cortica, Ltd.||System and method for identification of deviations from periodic behavior patterns in multimedia content|
|US9449001||Nov 3, 2014||Sep 20, 2016||Cortica, Ltd.||System and method for generation of signatures for multimedia data elements|
|US9454603||Aug 6, 2010||Sep 27, 2016||International Business Machines Corporation||Semantically aware, dynamic, multi-modal concordance for unstructured information analysis|
|US9466068||Feb 13, 2015||Oct 11, 2016||Cortica, Ltd.||System and method for determining a pupillary response to a multimedia data element|
|US9477658||Jun 12, 2014||Oct 25, 2016||Cortica, Ltd.||Systems and method for speech to speech translation using cores of a natural liquid architecture system|
|US9529984||Oct 1, 2013||Dec 27, 2016||Cortica, Ltd.||System and method for verification of user identification based on multimedia content elements|
|US20060047632 *||Aug 12, 2004||Mar 2, 2006||Guoming Zhang||Method using ontology and user query processing to solve inventor problems and user problems|
|US20060053000 *||Oct 7, 2005||Mar 9, 2006||Moldovan Dan I||Natural language question answering system and method utilizing multi-modal logic|
|US20080208840 *||Sep 21, 2007||Aug 28, 2008||Microsoft Corporation||Diverse Topic Phrase Extraction|
|US20100032418 *||Sep 27, 2007||Feb 11, 2010||Hamamatsu Photonics K.K.||Laser processing method|
|US20100042646 *||Oct 21, 2009||Feb 18, 2010||Cortica, Ltd.||System and Methods Thereof for Generation of Searchable Structures Respective of Multimedia Data Content|
|US20100228744 *||Feb 24, 2009||Sep 9, 2010||Microsoft Corporation||Intelligent enhancement of a search result snippet|
|US20100262609 *||Jun 23, 2010||Oct 14, 2010||Cortica, Ltd.||System and method for linking multimedia data elements to web pages|
|US20110082860 *||Apr 30, 2010||Apr 7, 2011||Alibaba Group Holding Limited||Search Method, Apparatus and System|
|US20120131073 *||Nov 19, 2010||May 24, 2012||Olney Andrew Mcgregor||System and method for automatic extraction of conceptual graphs|
|US20140040269 *||Oct 22, 2013||Feb 6, 2014||Ebay Inc.||Search clustering|
|US20140173400 *||Oct 23, 2013||Jun 19, 2014||Thomson Reuters Global Resources||Dynamic function builder|
|U.S. Classification||1/1, 707/E17.108, 707/999.003, 707/999.01, 707/999.002|
|International Classification||G06F17/30, G06F7/00|
|Cooperative Classification||Y10S707/99933, G06F17/30864, Y10S707/99932|
|Mar 16, 2007||AS||Assignment|
Owner name: INTEL CORPORATION, CALIFORNIA
Free format text: EMPLOYEE AGREEMENT WITH REDACTED SOCIAL SECURITY NUMBER;ASSIGNOR:ZHOU, JOE F.;REEL/FRAME:019024/0471
Effective date: 19990607
|Sep 21, 2011||FPAY||Fee payment|
Year of fee payment: 4
|Feb 5, 2016||REMI||Maintenance fee reminder mailed|
|Jun 24, 2016||LAPS||Lapse for failure to pay maintenance fees|
|Aug 16, 2016||FP||Expired due to failure to pay maintenance fee|
Effective date: 20160624